|
Post by dalenichol on Nov 15, 2020 21:44:16 GMT
jawzy and @dirk Nowitzki can I ask, even if briefly, how you guys are using the addition of xG? In what way is it benefiting you? That sounds more of a snippy question when it's really not intended to be, but I see people raving about it on twitter and I guess I don't really understand how I'm meant to interpret it?
|
|
|
Post by Dirk Nowitzki on Nov 15, 2020 22:35:00 GMT
xG is a metric that estimates the probability of scoring on a shot, based on historical data of maaaany shots. This takes into account factors like distance to goal, goalkeeper positioning, position of defenders, whether the shot was a header, and some other factors. I obviously don't know how the xG in FM is constructed, but since it's a game and they basically 'know' all variables exactly and can simulate loads of shots, it should be an accurate estimate for the game.
I'm not sure yet how the addition is really going to affect the way I play tbh, but it can certainly help to determine whether you have a problem or not. Football is a low scoring game where the outcome of a shot is either 0 or 1 goals. So let's say you have 5 shots with a xG of 0.2, you'd expect to score 1 goal, but you may get lucky and score 4 which makes it look like you played fantastically. This makes xG in single games not very useful, since there is so much variance and the actual goals you scored may often differ from what xG would make you expect. In the long run though, xG is a very good at predicting the amount of goals scored. There will be teams and players over/under performing their xG values, but in the long run this will even out, and there are hardly any players/teams that overperform xG season after season.
As for how to use this in FM; it can help to see patterns in performances I think. If the results are disappointing but the xG suggests you outplayed your opponents all the time, you may just have to keep going, try to improve morale and trust that the results will come as they go back to what you would expect based on xG (regression to the mean). On the other hand, if you are winning but they are all lucky wins you may want to change up your tactics anyway. You could already do this of course in older versions, but xG gives a more accurate idea of what is going on than the amount of shots or half chances and clear cut chances in previous versions. I wish they added a plot like this: Something like a running average of your last 10 games, this could help to see whether your changes to your tactic make an impact.
Like you had the problem with Morelos at Rangers, you can also see whether your attacking players come in good shooting positions or not. In the long run, getting into good scoring positions (high xG/90 minutes for example) but not converting the chances is less worying than massively outperforming xG, because that is likely a fluke and not sustainable.
For player recruitment, xG could be great but since we can just see the attributes of all players it is less impactful to have this additional information than it is in real life. There, all you have is obviously video and data, not a handy list of different skills with values between 1-20. xG data can then be used to identify strikers that come into great scoring position often for example, but there are more uses. Unfortunately we don't have this in FM, but data scientists in football also try to estimate xG created by each player on the pitch for a given shot. Like you could have xA (expected assists, the likelyhood of a pass that ends in a shot to be scored), but every pass leading to the assist also contributed to the quality of the chance. So basically the xG of a shot would be attributed to several players on the pitch. I'm more of a basketball nerd so I don't really know that much about this though, but I guess this is still an area where progress is being made because it's hard to measure. If done properly, it could really help to identify players that help creating good chances for your team.
Anyway, I rambled on a bit too much, so TLDR: You can more accurately spot if your struggle/glory is likely to persist, since xG is a good predictor of the actual goals you will score/concede in the longer run. On the individual player level, I'm not sure the use in FM goes much further than telling the same for offensive players (or whoever you want to score your goals).
I guess the main attraction for me is simply that I'm interested in the use of data in sports, and this is certainly a better tool than shots (on target).
|
|
|
Post by jawzy on Nov 16, 2020 1:22:28 GMT
I would use xG in terms of an extra tool in scouting
Dirk has essentially explained my take.
|
|
|
Post by dalenichol on Nov 16, 2020 9:08:11 GMT
Cheers for the responses guys, appreciate it. I understand the implementation then, but I just don't think it's going to greatly impact how I play the game. I'm back on my Rangers save for the beta, so we're expected to win and expected to score goals, which we're doing. It is interesting to see though, that the leagues top scorer is scoring almost double his xG (Eamon Brophy, 21 goals 11xG) but my scouts just don't rate him. That scoring record does make him stand out as an interesting player though, especially as someone that'd be so cheap.
|
|
|
Post by Dirk Nowitzki on Nov 16, 2020 11:09:35 GMT
The problem with this player is that he likely will not be able to keep this up. Almost nobody consistently overperforms xG. This is a common fallacy in player recruitment, which leads to overpaying for strikers that just went hot or lucky on their finishing. For example, a player scores 17 goals on 8 xG in 20 games, a team pays loads of money for him thinking he'll score a lot for them as well. Most often this will lead to disappointments because he will regress to the mean. For example, have a look at the xG stats for Robert Lewandowski since 2014. I guess he would be regarded as one of the most clinical finishers in the game, but he even underperformed xG since 2014. What makes good attacking players so good is that they get into good scoring positions often, accumulating a lot of xG, not that they finish much better than expected. I went trough some of the best players in the world, and probably the most extreme value I came across was Messi, who scored 204 goals on 170.95 xG. This is about 20% more goals than expected, but this is Messi, a ridiculous player. So I certainly wouldn't count on Eamon Brophy being able to keep up such crazy finishing rates that he outperforms xG by almost 100%. The caveat I'd like to make though is that this is a game, and it may function a bit different than real life. Maybe Brophy is some game breaking striker. ¯\_(ツ)_/¯ I don't really know how the AI valuates his own players, but if it were based on performances you might find better deals in players underperforming xG, trusting they will catch up at some point. I think it's more based on attributes and the underlying CA/PA though, so I'm afraid this doesn't really make too much sense in the context of the game. Edit: The information on the player pages I linked to is the kind of stuff I would loooove to be able to see in FM!
|
|
|
Post by jawzy on Nov 16, 2020 14:17:17 GMT
If I was to invest time into the FM analysing screens I could try and tweak tactics too based around xG, but i am bit to lazy to do that!
|
|
|
Post by jawzy on Nov 16, 2020 14:50:30 GMT
I love how Wingers actually make decent crosses into the box rather than shooting from a tight angle. combined with the new animations, it does give a lil wow factor everytime it happens
|
|
|
Post by dalenichol on Nov 17, 2020 10:49:30 GMT
|
|
|
Post by chaodck on Nov 17, 2020 13:17:27 GMT
That looks great indeed. Besides, it'll be great to take screenies for reports as well.
|
|
|
Post by Nick on Nov 17, 2020 16:58:48 GMT
Jesus, it's going to be impossible for me to not buy this, isn't it?
|
|
|
Post by Nick on Nov 17, 2020 16:59:57 GMT
Someone talk to me about the processing of the game. It's supposed to be quicker this year isn't it? Is that true?
Are people finding they can load more nations/leagues as a result?
|
|
|
Post by dalenichol on Nov 17, 2020 17:47:13 GMT
I mean, the issue is still the number of players and that's going to be what holds you back/determines how much you're willing to wait on processing times.
That said, yes, the processing times are much quicker and much smoother, and that includes saving and loading. I still autosave weekly out of habit from instable years gone by and I sometimes don't even notice that the games saving, it just runs so smoothly. Definitely no issues with processing
|
|
|
Post by chaodck on Nov 17, 2020 18:11:28 GMT
Jesus, it's going to be impossible for me to not buy this, isn't it? Busteeeeed!!! Great to have you back boss, I've missed you around here.
|
|
|
Post by Nick on Nov 17, 2020 18:22:06 GMT
Can't do anything on Steam without people seeing these days. First impression, your manager actually looks like a real person now. This is good. Second impression, how is Canada now an 'out of the box' league, but the rest of South America isn't?!
|
|
|
Post by Reiver on Nov 17, 2020 18:30:21 GMT
I was about to comment the same! Canada was already an "official" add-on last year. i think they managed to get it licensed so this year it became an out of the box league. I think being licensed is the key here.
|
|
|
Post by jawzy on Nov 17, 2020 19:04:04 GMT
Shame J league clubs lack of facilities screen gonna suffer again when custom db comes out
|
|
|
Post by dalenichol on Nov 17, 2020 19:05:39 GMT
Shame J league clubs lack of facilities screen gonna suffer again when custom db comes out Somebody managed to create a skin last year that fixed the Japanese issue. I would imagine that might come out pretty quickly this year.
|
|
|
Post by dalenichol on Nov 17, 2020 20:47:04 GMT
All these new player interaction options and you still cant:
A) Tell a player who's upset you sold somebody that "The player wanted to leave"
and
B) Tell a youth player that's unhappy you wont loan him out that "he's part of the first team now".
Like, what the fuck?
|
|
|
Post by Reiver on Nov 17, 2020 23:31:28 GMT
That is ridiculous. On the other hand, I like to see stuff like this: Now, is there a place to see this but for a specific player? EDIT - Silly me!
|
|
|
Post by jawzy on Nov 18, 2020 1:05:26 GMT
Sporar. id like to keep tabs on - simular level to your club, seems more clinical, I just have a dislike to people that end up with high amount of shots though.
|
|
|
Post by Nick on Nov 18, 2020 15:46:05 GMT
I'm a grumpy old man with this 'xG' stuff. I can see how the expected goals is a useful addition when it comes to scouting strikers, but that team xG screenshot that Reiver posted can get in the fucking bin. It just smacks of something SI include BeCaUsE fEaTuReS!!!! I'll honestly never use it.
Meanwhile, the two examples Dale mentioned slip through the net. It's the same every year.
|
|
|
Post by chaodck on Nov 18, 2020 16:59:59 GMT
I'm a grumpy old man with this 'xG' stuff. I can see how the expected goals is a useful addition when it comes to scouting strikers, but that team xG screenshot that Reiver posted can get in the fucking bin. It just smacks of something SI include BeCaUsE fEaTuReS!!!! I'll honestly never use it. Meanwhile, the two examples Dale mentioned slip through the net. It's the same every year. I think team xG can be used in two ways: 1.- To determine whether you need an overall tactical adjustment, due to generating more or less clear chances and also due to being more or less clinical (though the latter would probably merit a personnel change to improve). 2.- Used alongside next opponent scouting to try and adjust your tactics to their style. Defend more loosely against a low volume and wasteful team and put more emphasis on defensive work when a you face a more aggresive and clinical team. Number two sounds a bit too hands-on to my taste on paper, but I would've said the same about adjusting training calendars on a weekly basis and in FM 20 I ended up doing it for all 3 teams.
|
|
|
Post by Reiver on Nov 19, 2020 11:26:36 GMT
Another glaring option I want to be able to choose when talking to players and it just isn't there: when a player wants a new contract, i want to be able to tell him that there isn't budget for it right now.
|
|
|
Post by jawzy on Nov 19, 2020 14:18:13 GMT
I do like this screen though early in season i was more on the agressive clinical side, not quite how Liverpool have gone though! Also like this screen
|
|
|
Post by Reiver on Nov 19, 2020 14:46:47 GMT
I just saw one of my players not attempt to go for a ball because he realised he was offline, that was a nice touch.
|
|
|
Post by chaodck on Nov 19, 2020 15:14:35 GMT
I do like this screen That chart I was referring to when I mentioned about opponent scouting. At least on paper I find it super useful.
|
|
|
Post by Nick on Nov 19, 2020 16:53:26 GMT
Yeah, that attacking efficiency graph is more my kind of thing, I like that.
|
|
|
Post by jawzy on Nov 19, 2020 18:53:12 GMT
There is a defensive one too.
|
|
|
Post by richruzzian on Nov 20, 2020 12:58:39 GMT
I'm a grumpy old man with this 'xG' stuff. I can see how the expected goals is a useful addition when it comes to scouting strikers, but that team xG screenshot that Reiver posted can get in the fucking bin. It just smacks of something SI include BeCaUsE fEaTuReS!!!! I'll honestly never use it. Meanwhile, the two examples Dale mentioned slip through the net. It's the same every year. The real funny stuff is that XG stuuf OPta uses isn't in FM21. No SI decided to create there own version of it.
|
|
|
Post by Dirk Nowitzki on Nov 20, 2020 15:15:05 GMT
I do like this screen I feel like this screen would be so much more useful if it had xG per shot on the horizontal axis instead of the conversion rate. Now it's hard to distinguish whether you created good scoring opportunities or just finished very well, the former being much more sustainable and therefore more predictive of the future.
|
|